If one lidar hits another, it will result in at most one bad reading (perhaps a bad column?). This can likely be filtered, or a bad scan (360deg) can be altogether rejected and the data predicted using models based on past sensor readings.
I guess phase and timing sensitivity help a lot, because it's unlikely that another emitter will perfectly match your emission/detection duty cycle. It's also hard to get hundreds of cars at one intersection, because cars are very big.
The key terms in your literature/patent search should probably be "Crosstalk" and "multi-LIDaR".
There's no way a sensor can tell if a signal was from its own origin?
Guessing any signal should be treated as untrusted until verified somehow
but I suspect coders won't be doing that unless it's easy
The key terms in your literature/patent search should probably be "Crosstalk" and "multi-LIDaR".
E.g Livox mid 360 https://store.dji.com/en/product/livox-mid-360
https://github.com/PiLiDAR/PiLiDAR